dmitryikh / leaves

pure Go implementation of prediction part for GBRT (Gradient Boosting Regression Trees) models from popular frameworks
MIT License
419 stars 72 forks source link

adds ability to support multi:softprob for xgboost #74

Closed kgwinnup closed 4 years ago

kgwinnup commented 4 years ago

allows for use of multi:softprob for xgboost by defining a Softprob transformation and making the appropriate registration in transformation.go. Lastly, adds the transformation loading check for "multi:softprob" when loading an xgboost model.

coveralls commented 4 years ago

Pull Request Test Coverage Report for Build 156


Changes Missing Coverage Covered Lines Changed/Added Lines %
xgensemble_io.go 1 2 50.0%
transformation/softprob.go 0 24 0.0%
<!-- Total: 3 28 10.71% -->
Totals Coverage Status
Change from base Build 155: -0.9%
Covered Lines: 1674
Relevant Lines: 2418

💛 - Coveralls